2
$\begingroup$

I was wondering if geodesics are defined for all time on compact Finsler manifolds, or more generally, for any spray on a compact manifold (where by geodesics, I simply mean the integral curves of the spray).

I thought perhaps that the homogeneity condition of a spray would mean that the "size" of vectors couldn't grow too quickly along integral curves. To give a meaning to "size," I thought it would be useful to endow $TM$ with a complete metric, and the natural choice seemed to be the Sasaki metric. But then I realized, I didn't actually know if compactness of $M$ guarantees that the Sasaki metric is complete (I now know that it is). So I ask the following three sub-questions:

  1. If $(M,g)$ is complete, is the Sasaki metric on $TM$ necessarily complete?
  2. If $M$ is compact, is every spray on $M$ a complete vector field on $TM$?
  3. If $M$ is compact and Finsler, do geodesics on $M$ exist for all time?
$\endgroup$
10
  • 4
    $\begingroup$ Regarding (3). Yes, they do. Consider the geodesic flow as a Hamiltonian system on the cotangent bundle, the energy levels, which are invariant under local flow, are compact and so the geodesic exist for all time. $\endgroup$ Commented May 10, 2021 at 8:44
  • 1
    $\begingroup$ My dual girth paper has very little Finsler geometry. If you want background for it, it is best to take up Arnold and Givental's really nice survey on symplectic geometry in link.springer.com/chapter/10.1007/978-3-662-06791-8_1 $\endgroup$ Commented May 10, 2021 at 21:08
  • 1
    $\begingroup$ While you are at it check out Dmitry Faifman's paper: projecteuclid.org/journals/journal-of-differential-geometry/… $\endgroup$ Commented May 10, 2021 at 21:10
  • 1
    $\begingroup$ Both papers use the same background and make a nice pair, if I say so myself. $\endgroup$ Commented May 10, 2021 at 21:11
  • 1
    $\begingroup$ For a quick review of all of those things you can go to the survey sciencedirect.com/science/article/pii/S187457410680004X you can also find the pdf by googling if you don't have access. $\endgroup$ Commented May 11, 2021 at 10:11

1 Answer 1

1
$\begingroup$

These answers are primarily due to Juan Carlos Álvarez-Paiva's comments. I'm just recounting them:

  1. If $(M,g)$ is a complete, then $TM$ is also complete under the Sasaki metric, as shown here.

  2. Not every spray over a compact manifold is complete. Álvarez-Paiva linked this paper, which gives (as part of a more detailed study) an example of a Lorentzian metric on a 2-torus, where geodesics are not defined for all time. The idea is that a closed geodesic $\gamma:[0,1]\rightarrow M$ may increase in speed: $$\gamma(1)=\gamma(0)\quad\text{and}\quad\lambda\cdot \gamma'(1)=\gamma'(0)\quad\text{with}\quad\lambda<1.$$ Then $\gamma$ extends to go around infinitely many times and on the $n^\text{th}$ time around, the initial speed is $\lambda^{1-n}\gamma'(0)$ and the orbit takes time $\lambda^{n-1}$. Hence, the first $n$ times around takes time $$\sum_{k=0}^{n-1}\lambda^{k}=\frac{1-\lambda^{n}}{1-\lambda}.$$ As $n\rightarrow \infty$, this time converges to a finite number $\frac{1}{1-\lambda}$, which is the furthest that $\gamma$ extends.

  3. Again pointed out by Álvarez-Paiva: on a compact Finsler manifold, the sphere bundles are compact and the geodesic spray is tangent to these sphere bundles. Therefore, in this case, geodesic spray is complete!

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .